For a smartphone or a digital camera capable of capturing three-dimensional (3D) images, it usually takes two images of a same scene, in which the two images have a correct relative order and a parallax between them, and then arranges the two images into a 3D image. In addition, a user may upload two-dimensional (2D) images and utilize a 3D image creating software to generate corresponding 3D images of the uploaded 2D images. The user may further generate a 3D video based on multiple 3D video frames (i.e. 3D images), which can be played back on a 3D video player.
However, the user may not notice the relative order between the two images forming the 3D image or 3D video frame, for example, a left eye view and a right eye view, and thus may arrange the two images in an incorrect order, resulting an incorrect 3D image. For example, the user's left eye may see the right eye view and the user's right eye may see the left eye view, causing 3D image depth to be reversed to the user. That is, a pseudoscopic image or a pseudoscopic view may be generated and the viewing experience may be affected.
According to the present disclosure, detection and correction of the pseudoscopic image is of significant importance. If the pseudoscopic image cannot be detected and corrected, the user experience will be significantly degraded. To detect whether there is a pseudoscopy between a left view and a right view may often need to utilize corners respectively detected in the left view and the right view.
The disclosed methods and systems are directed to solve one or more problems set forth above and other problems.